Re: [PATCH v2] Input: soc_button_array: Remove kfree on data allocated with devm_zalloc

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

On July 27, 2014 11:50:41 PM PDT, pramod gurav <pramod.gurav.etc@gmail.com> wrote:
>Hi Dmitry,
>
>Thanks for the review.
>
>On Fri, Jul 25, 2014 at 9:52 PM, Dmitry Torokhov
><dmitry.torokhov@gmail.com> wrote:
>> Hi Pramod,
>>
>> On Fri, Jul 25, 2014 at 05:04:34PM +0530, pramod.gurav.etc@gmail.com
>wrote:
>>> From: Pramod Gurav <pramod.gurav@smartplayin.com>
>>>
>>> This patch does below:
>>> - Removes kfree done on data allocated with devm_zalloc in probe
>>>   path of the driver.
>>> - Adds a check on return value from devm_kzalloc which was missing
>>>
>>> CC: Dmitry Torokhov <dmitry.torokhov@gmail.com>
>>> CC: Lejun Zhu <lejun.zhu@linux.intel.com>
>>> CC: Sachin Kamat <sachin.kamat@linaro.org>
>>>
>>> Signed-off-by: Pramod Gurav <pramod.gurav@smartplayin.com>
>>> ---
>>>  drivers/input/misc/soc_button_array.c |   17 +++++++----------
>>>  1 file changed, 7 insertions(+), 10 deletions(-)
>>>
>>> diff --git a/drivers/input/misc/soc_button_array.c
>b/drivers/input/misc/soc_button_array.c
>>> index 5a6334b..fc64ec6 100644
>>> --- a/drivers/input/misc/soc_button_array.c
>>> +++ b/drivers/input/misc/soc_button_array.c
>>> @@ -83,6 +83,9 @@ soc_button_device_create(struct pnp_dev *pdev,
>>>                                      sizeof(*gpio_keys_pdata) +
>>>                                       sizeof(*gpio_keys) *
>MAX_NBUTTONS,
>>>                                      GFP_KERNEL);
>>> +     if (!gpio_keys_pdata)
>>> +             return ERR_PTR(-ENOMEM);
>>
>> OK, that makes sense.
>>
>>> +
>>>       gpio_keys = (void *)(gpio_keys_pdata + 1);
>>>
>>>       for (info = button_info; info->name; info++) {
>>> @@ -102,20 +105,16 @@ soc_button_device_create(struct pnp_dev *pdev,
>>>               n_buttons++;
>>>       }
>>>
>>> -     if (n_buttons == 0) {
>>> -             error = -ENODEV;
>>> -             goto err_free_mem;
>>> -     }
>>> +     if (n_buttons == 0)
>>> +             return ERR_PTR(-ENODEV);
>>
>> But that one and the rest don't, because failure in
>> soc_button_device_create() does not necessarily mean that binding for
>> the whole device will fail. In this case we do not want unused memory
>> hang around.
>Agree. Should resend the patch with only the error check after mem
>allocation and will be little more careful while sending any such
>change. :)

No need to resend, I picked out the good bits and applied.


Thanks.

-- 
Dmitry